WO2020002863A1

WO2020002863A1 - Method and apparatus for radar imaging

Info

Publication number: WO2020002863A1
Application number: PCT/GB2019/000085
Authority: WO
Inventors: Gordon Vigurs
Original assignee: The Secretary Of State For Defence
Priority date: 2018-06-29
Filing date: 2019-06-19
Publication date: 2020-01-02
Also published as: GB201908752D0; GB2576238B; GB2576238A

Abstract

A method and apparatus for generating a radar image. The method has the steps of A) Generating radar measurements using a radar signal, B) Performing interferometry on groups of the radar measurements to generate Skyline images, where the number of frequencies and/or number of measurement locations is sufficient to resolve scatterer location ambiguity, and C) Generating an image by evaluating variations of values in corresponding locations of multiple Skyline images based on non-parallel baselines, to identify a scatterer intensity at various positions.

Description

Method and Apparatus for Radar Imaging

The present invention relates to the field of radar imaging, particularly to imaging of a terrain.

In the past, the main methods of radar imaging included from a fixed location (E.g. using a rotating dish or a phased array antenna), or by moving a radar device along a path, known as synthetic aperture (SAR).

Of the two approaches, SAR approach is most useful for imaging, but monostatic (i.e.

transmitting and receiving a signal at the same location) SAR generally causes layover (layover makes tall objects appear to be fallen over and hilly areas distorted) and shadowing (a shadow to the side of a tall object). This is undesirable in a map, especially in a map for a lay audience.

It is known that measurement from a very shallow or zero grazing angle alleviates layover but increases shadowing (causing extreme shadowing as the measurement angle approaches horizontal), whilst using a higher measurement angle alleviates shadowing but increases layover.

Both layover and shadowing can be mitigated, using advanced multistatic SAR with appropriately advanced data processing techniques, however this requires measurements to be performed from a great diversity of directions at long range, and since shallow viewing directions are still needed, and these necessitate longer range measurements, arranging the viewing platforms is particularly expensive.

It is an object of the invention to provide a method and apparatus for generating an image reducing at least one of the above mentioned problems.

A first aspect of the present invention is set out in claim 1.

A second aspect of the present invention is set out in claim 2. A third aspect of the present invention is set out in claim 19.

A fourth aspect of the present invention is set out in claim 20.

A fifth aspect of the present invention is set out in claim 21.

A sixth aspect of the present invention is set out in claim 22.

Thus generally an image can be generated, by comparing the signal phases obtained at pairs of positions either end of baselines. The phase differences are calculated for a range of frequencies (or at one or more frequencies using multiple pairs that have parallel baselines but different baseline lengths). This provides a set of interferometers. By calculating the phase difference over the frequency range and/or using multiple length baseline

measurements, the ambiguity characteristic of interferometry is reduced (generally removed). This processing is generally performed within each range cell (i.e. typically each height), producing an image resembling a skyline as if viewed from a direction perpendicular to the interferometer baseline (i.e. similar to an observer's view on the terrain and looking across the terrain). This interferometric image is referred to herein for simplicity as a 'Skyline' image. Note that whilst with SAR, range is usually considered to be a direction along a terrain, here the term 'range' typically means a substantially vertical direction, since this is generally a more downward-looking approach.

The interferometric Skyline images are combined into an output (e.g. complete 3D image) by a processing step that takes into account how the complex values (or more accurately, at least their real components) vary between the Skyline images. However note that an interferometry measurement producing a skyline image generally results in (an array of) values which are complex in the sense of having real and imaginary parts due to being the mathematical result of operations involving complex numbers, however generally the imaginary part is insignificant meaning that these complex values are equivalent to real values, and thus it makes no practical difference whether they are recorded in a format that explicitly identifies the negligible imaginary components.

Various techniques such as tomographic processing may be used depending on the geometric arrangement of the interferometry baselines. This enables an image (typically a 3D image) to be built up, by use of baselines at varying angles along and through the terrain. The population of complex value 'skylines' are correlated against the expected coherent returns from terrain scatterers to determine the brightness of (typically all) voxels of the terrain area. The 3D image can then be processed in various ways, such as flattening it to form a 2D image that substantially lacks layover. Or other data, such as topography, can be extracted instead of generating a 3D image.

Note that the frequency range may be contiguous or non-contiguous, or may be multiple contiguous ranges, and the or each contiguous range may be a series of a large number of discrete frequencies. The radar image may be a 3D radar image, or may be a flattened view (typically vertically flattened) of the intensity of scatter determined in 3D. The radar image can be any form of image but preferably is a digital image. It is typically a digital 2D image but can equally be a physically printed 2D image, or a 3D digital model or a 3D printed model. Whilst comparing variations of complex numbers (or at least the real components thereof) in Skyline images to determine intensity of scatter at each position can be performed iteratively and individually for each position, it should generally be performed for all positions as one operation since this will typically require vastly less computer resources, energy and time (This can be done by appropriate use of a Fourier Transform, or by other equivalent or relevant algorithms). The term 'complex number' refers to a value identifying both real and imaginary components (one or both of the components can sometimes be zero though), such as for example a pair of I and Q components.

The baseline should be a synthetic baseline - meaning that the interferometry measurement is performed via a synthetic interferometry baseline. This means that the two

measurements generally need to be performed either by a single platform moved between two locations to perform the two measurements, or need to be performed using two separate platforms (e.g. aircraft) arranged at the different locations. For terrain imaging, it will generally not be possible to arrange the two measurement locations within one platform (e.g. aircraft) at a single point in time, since this would require an impractically large platform (e.g. aircraft).

An aircraft is generally envisaged as the/each platform (e.g. a manned or autonomous airplane, but potentially a helicopter, aerial drone, quad/multi-copter etc), but optionally a satellite could be used as a platform, or for particular applications, even a crane, vehicle on a hillside etc)

The term 'height' typically relates to a height within a height range of interest, which typically is divided into height slices. The term height generally relates to the vertical axis, but an alternative imaging axis could be selected, for example if it is desired to generate an image that is not viewing the terrain from directly above. In step A the range is preferably substantially height (e.g. range measured in the direction downward). The term 'terrain' means a region of the earth's surface of desired to be imaged, this could be an area of a body of water such as the ocean, or could be a geographical area such as a mountain range or a city, or a smaller area such as a building.

Preferably the average measurement angle of each of the measurement locations relative to a centre of the terrain is in the range from 20 degrees to vertical, preferably in the range from 30 degrees to vertical, more preferably in the range 45 degrees to vertical, yet more preferably in the range 60 degrees to vertical, more preferably in the range 70 degrees to vertical.

The invention particularly suits the use of higher and indeed very steep measurement angles, which permits the measurement locations to be much closer together, which in turn dramatically reduces a path length required for a radar device to travel to those locations, which dramatically reduces the cost and/or delay, compared to conventional advanced multistatic SAR. High measurement angles also dramatically reduces the measurement distances involved when measurements need to be made from a predetermined minimum altitude (compared to grazing angles necessary with SAR), and the shorter measurement distance facilitates use of higher frequencies (in bands such as Ka and above, e.g. Ka/Kw, that are more strongly attenuated by air), and as a result, power requirements are reduced higher image resolution can be achieved.

The following radar techniques are mentioned purely for avoidance of confusion, and are not considered relevant to the present invention: One known radar measurement approach is known as "Interferometry SAR" (InSAR). In this approach, two images created, with differing phase centers, enabling height information to be obtained across the image. InSAR therefore involves performing interferometry on SAR image data. Furthermore it is believed that in the field of radar terrain imaging, it is not known to use interferometry, other than to augment SAR terrain imagery with terrain height estimation or with change detection.

Note that whilst InSAR involves doing SAR twice, getting two images, and then performing interferometry to identify the difference between the two images (rather than to image the terrain), the present invention involves performing multiple (e.g. many) phase coded radar measurements and processing them in pairs to generate multiple (e.g. many) respective interferometry measurements, and then

using/combining those interferometry measurements to generate an image. Thus it can be seen that InSAR is a technique that is unrelated to the present invention. Another known radar detection method is known as Radar Interferometry, which is not even related to radar imaging. This involves transmitting a single frequency towards a point-like target, and measuring a phase difference in the response received back at two receivers. This identifies the direction of the point-like target. This type of radar technique is not an imaging method, since it does not identify the scatterer strength at a variety of directions, but rather identifies the direction of the single dominant scatterer. Using a single frequency causes ambiguity in the direction, which can be solved in various ways, such as adding more antenna elements, or combining the techniques of monopulse and interferometry. Another known radar method is called "along track interferometry". This is a technique to resolve the ambiguity arising from radial scatterer motion in SAR images, and thus is similarly unrelated to the present invention. Also note that the interferometry step in the present invention should not be confused with monopulse processing. Monopulse processing used in isolation would only be suitable for identifying the location of a scatterer on the assumption that there is only one scatterer in the region of interest. It is proposed in the present invention to use interferometry with a phase coding scheme (such as a sweep across a frequency range), using a number of frequency increments dictated by the desired image width, to achieve or approach a desired theoretical resolution that is dictated by a bandwidth and baseline length, in order to resolve all scatterers in the desired image width.

For the avoidance of doubt, any radar measurement(s) which can be used to generate a skyline image can be used. This generally requires two phase-coded radar measurements performed from different locations or a plurality of non-phase-coded (i.e. single frequency) radar measurements at distributed locations in the direction of the interferometry baseline. Each measurement is generally a monostatic measurement, but if the resulting ambiguities were corrected for (e.g. with additional measurements), it is possible that bistatic measurements could be used. Generally two transmissions and two reception actions are necessary, however if the resulting ambiguities were corrected for (E.g. with additional measurements) it may be possible to transmit from only one location (and record via either two locations or two paths) or alternatively to receive from only one location (and transmit via either two locations or two paths). Typically however the simplest arrangement is two individual monostatic phase-coded measurements, that are combined to form the skyline image.

The ideal point spread function, characterizing the potential resolution is given by:

Where y is the distance between two resolvable scatterers. The parameter b is given by:

Where F is the bandwidth of the frequency range over which the interferometric measurements are made. The lower case f is the carrier frequency, d is the baseline of the particular interferometer, l is the radar wavelength and H is the height of the platform above the scene centre. Resolution therefore depends of the fractional bandwidth and the baseline length in wavelengths.

The swath width is limited by the resolvable increment in frequency between

interferometer measurements. This defines the Doppler resolution requirements:

Where Y is the swath width avoiding fold-over due to under-sampling, 'c' is the speed of light, and AF is the required frequency (Doppler) resolution.

Specific systems would be designed based on the constraints characterized by these formulas.

Optionally a plurality of measurement pairs have baselines that all intersect at a common point, and the step of Identifying the scatterer intensity at each location in the terrain comprises performing axial tomography on the plurality of respective Skyline images, about the common point. This has the advantage that existing algorithms or existing approaches can be more readily used.

Another alternative is other, or indeed arbitrary locations of the measurement pairs. In this case for each location in the terrain image that is to be generated, the relevant location in each Skyline image of each measurement pair is identified, i.e. the location which corresponds to the location in the terrain image. Based on a plurality (many) Skyline images at varying distances, the complex numbers can be used at each frequency to determine the scatterer intensity at that location in the terrain image. In general, a plurality of non-parallel baselines should be used, preferably at a range of angles that are distributed about 180 degrees, which since a baseline does not have a preferred direction (positive or negative) consequently means the baseline angles are also distributed about 360 degrees. Preferably the distribution of baseline angles is substantially free of substantial gaps, preferably with no gap greater than 30 degrees, more preferably no greater than 5 degrees, more preferably no greater than 1 degree more preferably no greater than 0.1 degree. The baselines do not need to intersect but it is helpful if they do since this can allow use of simpler algorithms such as axial tomography. If the baselines do not intersect, it may still be possible to use axial tomography (especially if the

interferometry measurements were nontheless of the same area). Other algorithms may be preferable however, and in the absence of other algorithms, the method of explicit correlation should be used. Explicit correlation is less efficient since it involves evaluating the intensity of each pixel (or voxel) separately (which can be achieved by correlating the voxel's notional scatterer intensity with each value from each skyline image).

The term 'non-parallel baselines' generally requires at least that the differing baseline directions should differ around an axis of a look/viewing direction, i.e. at different angles in a plane of an image or imaging surface or e.g. terrain being imaged. It is not necessary, but may be convenient for the respective pairs/groups of respective interferometry

measurements to all be in respective parallel planes (or indeed for them to all be in a common plane), however this arrangement can facilitate easier, simpler or faster data processing.

For further information on the principles and practical implementation of axial tomography to datasets, ample literature is readily available, such "Tomographic Image Reconstruction. An Introduction." Milan Zvolsky, and "Easy implementation of advanced tomography algorithms using the ASTRA toolbox with Spot operators" Folkert Bleichrodt et al Springer Science+Business Media New York 2015.

Generally the measurement locations are airborne locations and the measurements are obtained using one or more airborne platforms. The use of an airborne platform(s) is important because the array of locations where measurements are needed do not all lie on a single straight line, and so this approach would be challenging using satellites or ground positions on overlooking hills. Alternatively the measurements may be obtained using a ground-based receiver, such as a ground vehicle.

Preferably the measurement locations are above flight level 600. This has the advantage of reducing the air certification steps required, which reduces the cost and makes it more suitable for use by commercial mapping organisations.

Preferably the or each airborne platform is an autonomous aeroplane. This has the advantage of enabling controllable, repeatable and fast collection at each of the measurement locations.

Preferably the or each airborne platform is an airborne repeater arranged to repeat to a ground based receiver and/or from a ground based transmitter. This has the advantage of reducing the amount of equipment that is required on the airborne platform, reducing both weight, power requirement, and air certification requirements, which reduces the cost.

Optionally the number of airborne platforms is at least three. Preferably the number of airborne platforms is no more than two, preferable only one. Whilst with conventional advanced multistatic SAR it may be necessary to use several SAR devices, use a fast jet, or wait a long time, the present invention suits collection at locations that are much closer together, enabling one or two airborne radar devices to collect the necessary data in a shorter amount of time. Whilst the measurement locations are not all arranged in a straight line as with conventional monostatic SAR, they may be in a circle/polygon or in two parallel lines, enabling one or two radar devices to perform the measurements in a short amount of time. Alternatively, since shorter horizontal distances are involved, it becomes possible to use three or more radar devices, optionally one per measurement location, so that the image is collected in a very short period of time.

Optionally the measurement locations are along lines. Two substantially parallel lines may be used, but preferably the lines are not all parallel with each other. The lines may be in pairs of substantially parallel lines, with at least one pair being non-parallel with respect to at least one other pair. An example format is for two substantially parallel lines in one direction (e.g. along (above) the terrain), and two substantially parallel lines in a

substantially orthogonal direction (e.g. along (above) the terrain in a different direction).

The lines can be straight, but may also be curved, for example expressing substantially matching sinusoids. One benefit of substantially parallel lines is that they facilitate a lawn- mower style pattern (alternately passing back and forth along a terrain, whilst incrementally moving sideways along the terrain before each pass) which makes efficient use of an aircraft platform supporting the radar transmitter and/or receiver. The paths need to be neither exactly straight nor exactly parallel paths to achieve such efficiency. Preferably an airborne platform is controlled to move in at least one lawn-mower style pattern (alternately passing back and forth along a terrain, whilst incrementally moving sideways along the terrain before each pass). Optionally the (or at least one, optionally multiple) airborne platforms are controlled to move in at least two lawn-mower style patterns, preferably exactly two, optionally three. The back and forth directions of the patterns are different, for example the first lawn-mower style pattern being substantially orthogonal to the second, or if there are three then there may be a substantially l/6^th of a turn angular difference between them (in this context where the patterns are substantially reversible, the term l/6^th thus also covers 4/6^th).

Optionally Generating a radar image comprises Generating a radar image of the terrain of multiple height slices, and generating a 3D image or a flattened 2D image. This has the advantage of enabling inspection of data associated with a 3D volume without layover effects. If a 3D image is generated this enables inspection in 3D, for example creating a virtual model walkthrough or a 3D printed map. If a 2D image is generated, this enables sharing as a conventional image, whilst more accurately capturing vertical details such as buildings and wall boundaries, which is difficult in conventional monostatic SAR due to layover effects.

Optionally the measurement locations are clustered into one or more paths. Suitable examples include substantially circular, and substantially as two parallel lines. Many others are possible such as concentric circles, rectangles, and indeed any arbitrary set of points can be used. However the advantage of clustering them into one or more paths, is to enable one or more radar devices on airborne platforms to travel to and perform measurements at each of the locations in a shorter amount of time than if using non-path-clustered set of measurement locations.

Preferably, the radar measurement is performed at a frequency of Ka Band or above, preferably Kw Band. Higher frequencies generally enable higher bandwidths and thus enable higher image resolution images. In addition since high frequency bands such as Ka and Kw bands have high attenuation in the atmosphere, their use is synergistic with the use of an airborne platform to take the measurements and also synergistic with higher measurement angles than are conventional with for example SAR (i.e. above 20 degrees rather than around 10 degrees).

Any form of range processing may be used. Rather than compressing the pulse across the modulation, each frequency sample is generally processed individually so that each range measurement has an associated frequency sweep. Pulse compression is generally not used, since this affects the frequency sweep which should be retained for the interferometry. The result is to obtain I and Q values for each range and frequency sample.

Generally, the phase coding (if used as opposed to multiple measurement locations per baseline) in each radar measurement covers a frequency range sufficient to resolve ambiguity in interferometer angle measurement (either in isolation, or if desired, in combination with any additional techniques used to help resolve ambiguity) in the area of the terrain of interest. The range of frequencies may optionally be transmitted as a frequency chirp. Phase coding (i.e. frequency variation within a pulse) is generally necessary since conventional single-frequency radar interferometry is not sufficient. The use of a frequency range or other phase coding enables more information to be gathered regarding the terrain, to be able to feed into the later steps. Preferably a wide bandwidth is used such as at least 1MHz, preferably at least 1GHz. This enables higher resolution compared to a small bandwidth (in the direction of any particular baseline). The required frequency range is calculable from the baseline and required resolution in a manner which the person skilled in the art will be able to perform - see the formulas on the preceding pages. Typically step A comprises performing a radar measurement at each of a plurality of measurement locations that are above a terrain, and that are arranged in distributed fashion with respect to two horizontal dimensions. Typically step A comprises performing each radar measurement using in each case a signal that comprises phase-coding over a frequency range. Typically step A comprises generating respective radar measurements each comprising complex values as a function of frequency and range.

Typically step B comprises selecting pairs of the measurement locations, each pair defining a respective baseline, and for each pair of measurement locations, generating a Skyline image comprising complex values (which will each have a zero or substantially zero imaginary component), typically, for a plurality of heights and baseline-wise locations. Typically this is done by: For each of a plurality of range cells, correlating the phase of the respective radar measurements at each frequency in the frequency range against expected values for notional scatterers in each of a plurality of resolution cells with respect to the baseline, to generate an intensity for each resolution cell as a complex value (having variation in the real component);

Typically step C comprises, with respect to each of a plurality of positions in the terrain, evaluating variations of the complex values (at least of the real component) in

corresponding locations (e.g. resolution cells) of a plurality of the Skyline images, in each case taking into account the distances of the corresponding locations from the position, to identify a scatterer intensity at that position.

For the avoidance of doubt, the term 'interferometry measurement' requires interferometry to be performed, and thus is distinguished from for example the method known as synthetic aperture radar. The step of performing interferometry does not involve performing synthetic aperture radar, although it is possible for a user to perform synthetic aperture radar in addition to the present invention, and the resulting images from each could conceivably be combined or compared.

Further embodiments are set out in the claims. A preferred embodiment of the invention will now be described with reference to the figures in which:

Figure 1 shows an illustration of how SAR imaging works, according to the prior art; Figure 2 shows an illustration of how InSAR works, according to the prior art;

Figure 3 shows an illustration of step A of an embodiment of the present invention; Figure 4 shows an illustration of step B of an embodiment of the present invention; Figure 5 shows an illustration of step C of an embodiment of the present invention; Figure 6 shows an illustration of step B of an embodiment, showing interferometry being performed on a series of pairs of locations arranged in respective rows;

Figure 7 shows an illustration of a starting point for step C of an embodiment of the present invention, where multiple skyline images are processed;

Figure 8 shows an illustration of a possible output from step C of an embodiment of the present invention, where a 3D image is generated of a terrain with buildings; Figure 9 shows an illustration of a possible output from step C of an embodiment of the present invention, where a 2D map is generated of a terrain with buildings;

Figure 10 shows an example of an alternative configuration where each of the baselines intersect at a point, and the measurement locations are in a circle;

Figure 11 shows an alternative embodiment of the present invention where each interferometry measurement (only the first three are shown) and each resulting skyline image, results from combining data from multiple pairs of locations, rather than just one pair; and

Figure 12 shows an illustration of a simple example of a double lawn-mower style path as seen from above, for an airborne platform for performing measurements above a terrain, according to an embodiment of the invention.

Figure 1 shows a conventional approach to synthetic aperture radar (SAR) imaging (forward range from time delay, cross range from phase sequence over the integration period). An aircraft 1 passes a terrain 2. In monostatic SAR mode it transmits radar pulses towards the terrain 2 which are reflected back to be measured at the aircraft 1. Due to the relative movement of the aircraft 1, returned signals from a scatterer 3 will have different Doppler shift in different measurements, enabling it to be located when the signals are processed. The distance to the scatterer varies, but remains within range cell. Cross range position (y) determines the phase as a function of time. Measurements are taken along a linear path, to provide a synthetic imaging aperture.

Figure 2 shows another known radar method, known as Interferometry SAR (InSAR). In this approach conventional SAR imaging of a terrain 2 is performed twice via two aircraft passes, A and B. The two images are compared (taking into account the complex values, not merely the magnitude of scatterer intensity). For any given scatterer, 3, that is common to both images, the path length in image A is different from that in image B, and the phase difference provides a measure of the height of the scatterer 3. As a result height data is available for the terrain 2 being imaged. InSAr thus locates the height of a scatterer in a resolution cell common to two images generated by vertically separated phase centres. In summary, interferometry is performed on the SAR imagery, after the conventional SAR imaging has been performed twice.

Figure 3 shows an illustration of step A of an embodiment of the present invention. Here, an aircraft is shown above a terrain that contains buildings 4. The aircraft 1 is a platform for an interferometry measurement. Range is substantially in the vertical direction from the terrain 2. Range processing is performed for a sequence of pulses (typically having different frequencies), with sufficient bandwidth (or number of measurement locations along the baseline direction) required for interferometric processing. The platform is in the far field of the terrain (much higher than as drawn), thus the figure is not to scale. Forward range cells (horizontal lines shown) are determined by pulse delay in the returned signal.

Figure 4 shows two interferometry measurements being performed, thus forming an interferometry baseline between them. Using a single frequency, a scatterer can be located as being at one of a series of locations (dots).

Figure 5 shows how a second frequency can be used to limit the possible locations to a second set of possible locations (second row of dots), and since there is only one location where these locations intersect, this identifies the location of the scatterer. At the second frequency, all the 'ghost' scatterers are in different positions, and only the true scatterer remains in the same place. Use of more than two frequencies allows the ambiguity in the scatterer position to be removed from across a larger area. Typically a very large number of frequencies should be used. Alternatively or additionally more than two measurement locations that differ in position along the baseline direction can be used to eliminate such ambiguity.

Figure 6 shows an illustration of step B of an embodiment, showing interferometry measurements 5 being performed on a series of pairs of locations from respective platforms 1, with the pairs of locations being arranged as rows in respective lines. Each interferometry measurement collects information regarding the whole or part of the terrain, in this case an oval or circular area of the terrain (solid oval lines). As the aircraft 1 travels along its path (in this case a linear direction of travel is shown with arrow) it performs interferometry measurements sequentially along the path - initially each measurement is generally just a monostatic (optionally phase coded) measurement of the terrain. The aircraft then travels along the second line (or a second aircraft does so) performing the second set of (optionally phase-coded) measurements. The two sets of measurements may be deliberately aligned to form predetermined pairs, or alternatively pairs are later selected from the available measurement locations. It is not essential for each individual measurement to only be used in one interferometry pair.

The pairs of measurements are processed as an interferometry baseline (illustrated as solid double-ended arrows). This does not capture merely a thin 'slice' of the terrain, but rather captures data on an area of the terrain which overlaps strongly or completely with the areas associated with most or all of the other measurements. Each pair of measurements, processed as an interferometry measurement with respect to an interferometry baseline generates a respective "skyline" image, comprising an array of complex values (E.g. a complex vector, or real and imaginary values, also known as I and Q values).

As shown in figure 6, the pairs chosen result in baselines that are not parallel. Preferably an additional two lines of measurements are performed, and these are substantially orthogonal (55 to 125 degrees preferably 80 to 100 degrees), so as to form substantially a square or substantially # symbol shape or any variation therebetween. This facilitates pairs to be chosen to form baselines with angles with an angular distribution that exhibits gaps that are less wide than would typically be the case with only two straight lines, thus facilitating more similar maximum resolution to be achieved in all lateral directions (e.g. across the terrain).

Note that it is possible for the measurement orientation to remain fixed as the aircraft passes overhead, resulting in slightly differing measurement regions. This has the result that the output only relates to the area of the terrain where sufficient measurement areas intersect. As an example, measurements could be performed in a line along a large strip of terrain, and groups of those measurements are used to generate images, so that the whole strip of terrain can be imaged. Typically, however it is preferable for each measurement, and preferably each interferometric pair to gather data on the same or very similar region of terrain, so that all or substantially all of the measurements are useable in generating image data for the whole of the area of the terrain being imaged.

Figure 7 shows an illustration of a starting point for step C of an embodiment of the present invention, where a skyline images is shown. The skyline images, as generated, are composed of complex numbers and these complex numbers differ from image to image, however the imaginary components of the complex numbers will be zero, so they could be converted to real (non-complex) numbers and outputted as such (however further steps based on variations in those real numbers would nonetheless also be based on variations in the original compex numbers).

Figure 8 shows an illustration of a possible output from step C of an embodiment of the present invention, where a 3D image is generated of a terrain with buildings. By combining the data from the Skyline images (see figure 7) the location in 3D of each scatterer can be identified, thus generating a 3D image (model) 7.

Figure 9 shows an illustration of a possible output from step C of an embodiment of the present invention, where a 2D map is generated of a terrain with buildings 8, corresponding to a bird-eye-view of the 3D image or model as shown in figure 8. Figure 10 shows an example of an alternative configuration where each of the baselines intersect at a point, and the measurement locations are in a closed or open loop in one or two, or more, segments. In this example they are shown in a circle 9 however any loop in one or more segments could be used, such as an oval or a polygon comprised of substantially straight lines, such as a square/rectangle, a triangle or hexagon. The lines of the polygon optionally extend beyond a loop, such as a square/rectangle extending to form a # symbol shape.

Figure 11 shows an example of an embodiment where more than two measurements are used to generate each skyline image. In this case many pairs of measurements are used (pair A1 and Bl, pair A2 and B2, pair A3 and B3 etc), and the pairs have varying separations. The diversity of the separations provides a way to disambiguate scatterer location. This can be an alternative or an addition to using phase coding (e.g. swept frequency).

In this example the measurement locations are arranged on a path around a target area. A circular path is shown, but there is no requirement for it to be circular. To generate each skyline image, data from many pairs of locations are used together, and the pairs are chosen such that their respective individual baselines would be parallel, however they are used together to form a combined interferometry measurement with a single effective baseline.

To generate a second skyline image, with respect to a second effective baseline, this is repeated however the measurement locations are chosen and paired such as to provide many parallel individual baselines that are all parallel, but at a new angle (see second image in figure 11 - with all individual baselines shown at 5 degrees compared to the first example. These pairs of measurements are taken together to form a combined

interferometry measurement, and this results in a second skyline image.

Note that the same initial measurement gets used in the process of generating multiple (or many or all) of the skyline images, but in each case would be partnered with a different second measurement. For this reason it is beneficial for the arrangement of measurement locations to exhibit a high degree of rotational symmetry, e.g. 90-fold rotational symmetry or higher, preferably at least 360-fold rotational symmetry. The skyline images generated from the arrangement of pairs in the second and third arrangements shown in figure 11, have baselines arranged at 5 degrees and 10 degrees compared to the first arrangement. In practice many arrangements are used e.g. 90 or more, preferably 360 or more, preferably 1000 or more.

More generally: Each skyline image is generated, based on either:

1. a single pair of measurement locations with sufficient pulse coding (generally many frequencies) to resolve scatterer ambiguity,

2. measurements at a sufficient number of separations (generally many different

separations) sufficient to resolve scatterer ambiguity, or

3. a combination of pulse coding and measurement location separation diversity, that is sufficient to resolve scatterer ambiguity.

For the avoidance of doubt, note that baseline is not the same as the swath width (which is obtained from the Doppler resolution). Parameters defining the resolution are baseline length and the bandwidth of the frequency range employed.

Regarding the term 'skyline image' what is generated is the phase difference as a function of frequency of all scatterers in the antenna footprint. This is correlated against a unit scatterer response over the same frequency range, to generate the coherent sum of scatterers along stripes perpendicular the plane containing the beam footprint centre and the base line. This is applied in each forward range (e.g. height) cell, producing an image which is referred to as a 'skyline' as it is a plot of intensity vs forward range (e.g. height) and lateral range.

Repeating this process, either by changing the direction of the baseline, or translating it in space (or a mixture of both), between consecutive measurements, provides a time sequence for individual scatterers within these stripes. These should be resolved in the third dimension using standard SAR or tomography methods (or a hybrid of both). This final stage typically is to complete a 3D image of the region of interest. Lastly figure 12 shows an example of two orthogonal lawn-mower style paths (thick line) that one (or more, not shown) airborne platform(s) could follow to ensure that for any particular region (dotted loop), suitable baselines between measurements can be selected with variations in angle distributed such as to avoid significant angular gaps. Whilst only a small number of baselines (thin straight black line) are shown for each region, in practice typically many, typically hundreds or even thousands, would be used.

Two examples are shown in figure 12. The group of baselines in the example in the middle all crossing at a common point, and the baselines at the bottom right are non-parallel, but do not all cross at a common point. Many groups would typically be used, and typically they would overlap one another (not shown). The manner of how the data is combined depends on whether the group of interferometry baselines intersect at a point (in which case axial tomography can be used) or whether they are arbitrary non-parallel interferometry baselines. In both cases, at a minimum, it is possible to determine the intensity of scatter at each position iteratively and individually for each position within the image to be created. However it may be faster to combine all of the interferometry measurements and solve for them simultaneously to the extent possible using an optimisation algorithm.

As an additional aspect, there is provided a method of generating a radar image, comprising the steps of: A. Generating radar measurements, by: Performing a radar measurement at each of a plurality of measurement locations, and that are arranged in distributed fashion with respect to two dimensions that are perpendicular to a viewing direction, at at least one frequency, to generate respective radar measurements each comprising complex values; B. Performing interferometry to generate Skyline images, by: For each of a plurality of differing baselines: Selecting a group of measurement locations, which have differing positions along the baseline direction of the respective baseline; Performing interferometry on the respective radar measurements of the group, to generate a Skyline image by; for each of a plurality of resolution cells: Correlating the phase of each of the respective radar measurements of the group, at each of the at least one frequencies against expected values for a notional scatterer in each respective resolution cell, to generate an intensity for that resolution cell as a complex value; and Wherein, for each group: the number and diversity of positions of the measurement locations with respect to the baseline direction, and the number and diversity of frequencies; are sufficient to substantially avoid scatterer location ambiguity in generating the intensity for each of the plurality of resolution cells; C.

Generating an image, by: With respect to each of a plurality of positions, evaluating variations of the complex values in corresponding resolution cells of a plurality of the Skyline images, in each case taking into account the distances of the corresponding resolution cells on their respective baselines from that position, to identify a scatterer intensity at that position.

Claims

1. A method of generating a radar image, comprising the steps of:

A. Generating radar measurements, by:

Performing a radar measurement at each of a plurality of measurement locations, and that are arranged in distributed fashion with respect to two dimensions that are perpendicular to a viewing direction, at at least one frequency, to generate respective radar measurements each comprising complex values;

B. Performing interferometry to generate Skyline images, by:

For each of a plurality of non-parallel baselines:

Selecting a group of measurement locations, which have differing positions along the baseline direction of the respective baseline;

Performing interferometry on the respective radar measurements of the group, to generate a Skyline image by;

for each of a plurality of resolution cells:

Correlating the phase of each of the respective radar measurements of the group, at each of the at least one frequencies against expected values for a notional scatterer in each respective resolution cell, to generate an intensity for that resolution cell; and

Wherein, for each group:

the number and diversity of positions of the measurement locations with respect to the baseline direction, and

the number and diversity of frequencies;

are sufficient to substantially avoid scatterer location ambiguity in generating the intensity for each of the plurality of resolution cells;

C. Generating an image, by:

With respect to each of a plurality of positions, evaluating variations of the intensities in corresponding resolution cells of a plurality of the Skyline images, in each case taking into account the distances of the corresponding resolution cells on their respective baselines from that position, to identify a scatterer intensity at that position.

2. A computer implemented method of generating a radar image, comprising controlling a computer to perform the steps of:

A. Providing radar measurement relating to each of a plurality of measurement locations, and that are arranged in distributed fashion with respect to two dimensions that are perpendicular to a viewing direction, at at least one frequency, wherein the radar measurements each comprise complex values;

B. Performing interferometry to generate Skyline images, by:

For each of a plurality of non-parallel baselines:

for each of a plurality of resolution cells:

Wherein, for each group:

the number and diversity of positions of the measurement locations with respect to the baseline direction, andthe number and diversity of frequencies;

C. Generating an image, by:

With respect to each of a plurality of positions, evaluating variations of the intensities in corresponding resolution cells of a plurality of the Skyline images, in each case taking into account the distances of the corresponding resolution cells on their respective baselines, from that position, to identify a scatterer intensity at that position.

3. The method of any preceding claim, wherein the at least one frequency is a plurality of frequencies.

4. The method of any preceding claim, wherein in the step of Selecting a plurality of

measurement locations, in each case the plurality of measurement locations comprises at least three measurement locations, which have differing positions along the baseline direction of the respective baseline.

5. The method of any preceding claim, wherein step C comprises performing axial

tomography on the plurality of respective Skyline images, and preferably a plurality of measurement groups have baselines that all intersect at a common point.

6. The method of any preceding claim, wherein the baselines have respective angles that are distributed about 360 degrees, such as to avoid gaps greater than 30 degrees, preferably such as to avoid gaps greater than 1 degree.

7. A method according to any preceding claim, wherein the measurement locations are airborne locations and the measurements are obtained using one or more airborne platforms.

8. A method according to claim 7, wherein the measurements are obtained by flying at least one airborne platform to each of the measurement locations.

9. A method according to claim 7, wherein at least a plurality of the measurement

locations are respective locations of multiple antennas arranged on a single airborne platform or convoy.

10. A method according to claim 7, wherein the measurement locations are above flight level 600.

11. A method according to claim 7, wherein the or each airborne platform is an autonomous aircraft.

12. A method according to claim 7, wherein the or each airborne platform is an airborne repeater arranged to repeat to a ground based receiver and/or from a ground based transmitter.

13. A method according to claim 7, wherein the number of airborne platforms is no more than two, preferable only one.

14. The method of any preceding claim, comprising generating a radar image of multiple distance slices with respect to the viewing direction, and generating either a 3D image, or a 2D image thereof.

15. The method of any preceding claim, wherein the measurement locations are clustered along one or more paths, preferably one path, preferably a loop.

16. The method of any preceding claim, wherein each radar measurement is performed in a frequency band that is Ka Band or above, preferably Kw Band.

17. The method of any preceding claim, wherein phase coding in each radar measurement is provided as a frequency sweep with a complex value for each frequency and each range sample.

18. The method of any preceding claim, wherein the image is an image of a terrain, and wherein preferably the two directions perpendicular to the viewing direction are horizontal directions.

19. A data processing apparatus configured to perform step A, step B, and step C, as listed in claim 2 or any preceding claim dependent thereon.

20. A computer program comprising instructions which, when the computer program is executed by a computer, cause the computer to carry out steps A, B, and C, as listed in claim 2 or any claim dependent thereon.

21. A computer-readable medium comprising instructions which, when executed by a

computer, cause the computer to carry out steps A, B and C, as listed in claim 2 or any claim dependent thereon.

22. Apparatus for generating a radar image, comprising:

Radar measurement device arranged to:

A. Generate radar measurements, by:

Performing a radar measurement at each of a plurality of measurement locations, and that are arranged in distributed fashion with respect to two horizontal dimensions that are perpendicular to a viewing direction, at at least one frequency, to generate respective radar measurements each comprising complex values; and

At least one computer arranged to:

B. Perform interferometry to generate Skyline images, by:

For each of a plurality of non-parallel baselines:

for each of a plurality of resolution cells:

Wherein, for each group:

the number and diversity of frequencies;

are sufficient to substantially avoid scatterer location ambiguity in generating the intensity for each of the plurality of resolution cells; and

C. Generate an image by:

23. Apparatus according to claim 22, wherein the Radar Device comprises an airborne platform.

24. Apparatus according to claim 23, wherein the at least one frequency is a plurality of frequencies.

25. Apparatus according to claim 23, wherein the plurality of measurement locations comprises at least three measurement locations, which have differing positions along the baseline direction of the respective baseline.

26. Apparatus according to claim 22, wherein the at least one computer is arranged on the ground.